THE MOAT IS NOW A BRIDGE

And you get to choose who crosses it

For years, Microsoft's AI proposition rested on a single defensible truth: if your organisation runs on M365—Outlook, SharePoint, OneDrive, Teams—then Copilot was the only AI that could see your world clearly. That was the moat. Deep, wide, and expensive to cross from the outside.

Last week, Anthropic built a bridge.

Claude now connects natively to Outlook, SharePoint, and OneDrive—on every plan, including free. The moat didn't disappear. But it stopped being a barrier and became a throughway. The same M365 data, the same organisational context, now accessible to a different model entirely.

This isn't the end of Copilot. But it is the end of Copilot's most important argument.

𝗧𝗵𝗲𝗻 𝗰𝗮𝗺𝗲 𝘁𝗵𝗲 𝗯𝗶𝗹𝗹

Microsoft announced commercial M365 pricing increases of between 5% and 25% depending on plan, effective July 1, 2026. The E3 enterprise license now sits at $312 per user per year. E5—the plan most commonly sold on the strength of Copilot inclusion—reaches $720 per user per year.

The timing is instructive. Internal reports consistently suggest Copilot adoption in enterprise sits around 10-15% of licensed users against projections of 70%+. Microsoft is raising the price of a product most people aren't using, justified by AI features bundled into plans whether organisations want them or not.

One analysis put it plainly: when a platform already owns the workflow, a price rise can be both defensible and extractive simultaneously. Microsoft doesn't need every buyer to love the increase. It only needs most of them to decide that the cost and disruption of challenging the stack feels worse than absorbing it.

That calculation just became more interesting.

𝗦𝗼 𝘄𝗵𝗮𝘁 𝗱𝗼𝗲𝘀 𝘆𝗼𝘂𝗿 𝗼𝗿𝗴𝗮𝗻𝗶𝘀𝗮𝘁𝗶𝗼𝗻 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗰𝗵𝗼𝗼𝘀𝗲 𝗻𝗼𝘄?

There are three realistic positions and each comes with a different set of trade-offs.

MS only—Copilot as the single answer

The path of least resistance. One vendor, one procurement conversation, one integration story. Copilot sits inside the tools your people already use, which matters for adoption. The compliance and governance story is clean.

The honest question before July 1st: are your people actually using it?

If your organisation is running E5 to justify Copilot and your people aren't using it, July 1st is the most expensive way to keep doing nothing differently. The price increase lands on the same seat count. The adoption problem doesn't change with the invoice.

MS + OpenAI—The default inheritance

Many organisations are here already without having chosen it. Copilot runs on OpenAI models. If you have Copilot, you have OpenAI inside your M365 environment whether you decided that or not. The question is whether you're using it deliberately or just hosting it.

MS + Anthropic—The new option

This is the position that didn't exist a week ago and that procurement should be modelling now.

Claude connecting to M365 means you can run a meaningfully different model against your organisational data—one with a different capability profile, a different approach to safety and reliability, and a different commercial relationship.

The procurement logic: downgrade from E5 to E3. The saving at list price is $408 per user per year. Use a portion of that saving to fund Claude access. You keep the M365 infrastructure you've built around. You add genuine AI capability that your people will actually choose to use. You connect it to Outlook, SharePoint, and OneDrive through the new native connectors.

You come out ahead financially. You come out ahead on adoption. And you stop paying a premium for a feature that wasn't working.

This isn't anti-Microsoft. It's procurement doing its job.

𝗦𝗶𝗹𝘃𝗲𝗿 𝗯𝘂𝗹𝗹𝗲𝘁 𝗼𝗿 𝗰𝗼𝗳𝗳𝗶𝗻 𝗻𝗮𝗶𝗹 𝗳𝗼𝗿 𝗖𝗼𝗽𝗶𝗹𝗼𝘁?

Neither. And that framing misses what's actually happening.

The M365 connectors don't destroy Microsoft's AI position. Microsoft still owns the environment. The data still lives on their infrastructure. The governance, the compliance, the admin controls—all still Microsoft. For regulated industries that may remain decisive.

What has changed is the sentence Microsoft could previously say without challenge: to get real AI value from your M365 investment, you need Copilot. That sentence no longer holds. You need an AI that connects to M365. Copilot is one. It is no longer the only one.

𝗧𝗵𝗲 𝗽𝗮𝘁𝘁𝗲𝗿𝗻 𝗳𝗼𝗿𝗺𝗶𝗻𝗴

What we are watching—quietly, without fanfare—is the unbundling of the AI stack.

The model: what thinks. The data: what it thinks about. The platform: where it thinks. The interface: how you talk to it.

These four things are separating. Microsoft is the platform and the data. The model is increasingly a choice. Apple is making the same move—opening Siri to third party AI integration rather than competing in the model race. Both are betting that owning the environment is more durable than owning the intelligence that runs inside it. Platform over product. Infrastructure over model.

The organisations that understand this earliest will build AI stacks that are deliberately chosen rather than accidentally inherited—resilient to model changes, not locked to a single vendor's roadmap, and actually used by the people they are supposed to serve.

The moat is now a bridge. July 1st is a good time to decide what you're building on it.

This is the third in an ongoing series. Earlier posts: 

When Your Copilot Is Flying On Borrowed Fuel https://www.ssh-creative.com/post/99ae39db and

Copilot Switches Back To Manual Pilot https://www.ssh-creative.com/post/c96eaaeb

Sources: Microsoft 365 pricing announcement December 2025, effective July 1 2026. ManageEngine M365 pricing analysis. VaaSBlock Microsoft 365 price defence analysis. Enterprise Copilot adoption figures widely reported.